Hi all,
I’m repairing a late 60s AC coupled amplifier and I need to replace the output capacitors as one is leaky.
Being so old, there is endless space for the replacement of the original 1000uF, 35v units.
I know that larger capacitors give a lower roll off in the bass and if I have calculated it right the current 1000uF capacitor gives a f3 of 19hz with an 8 ohm load, which is fine. I accept then, that it doesn’t necessarily need to be sized larger, particularly as this is perhaps a 20w amp and deep bass isn’t really on the agenda anyway.
This means I can increase the voltage rating of the capacitor and have it still neatly fit the space on the PCB, is there any particular advantage to this in terms of distortion?
Specifically, would a 2200uF 50v capacitor be a better replacement or a 1500uF 80v in this case?
Is there an upper limit here, would 4700uf 100v be better than either of the above or would of fry the output transistors on start up?
I’m repairing a late 60s AC coupled amplifier and I need to replace the output capacitors as one is leaky.
Being so old, there is endless space for the replacement of the original 1000uF, 35v units.
I know that larger capacitors give a lower roll off in the bass and if I have calculated it right the current 1000uF capacitor gives a f3 of 19hz with an 8 ohm load, which is fine. I accept then, that it doesn’t necessarily need to be sized larger, particularly as this is perhaps a 20w amp and deep bass isn’t really on the agenda anyway.
This means I can increase the voltage rating of the capacitor and have it still neatly fit the space on the PCB, is there any particular advantage to this in terms of distortion?
Specifically, would a 2200uF 50v capacitor be a better replacement or a 1500uF 80v in this case?
Is there an upper limit here, would 4700uf 100v be better than either of the above or would of fry the output transistors on start up?
No fear, all the space in the world.
I’ve actually got this irrational urge to hot glue the original components to the steel chassis, after replacement, for future owners to marvel at...
The main PS filter cap is on the left, 1500uF at 63v and the two output caps on the right.
I could get a 10,000uF 100v PS filter cap in there at the same diameter and to the height of the case and who knows what I’m the output cap section.
I’ve limited myself to 4700uF for the PS in deference to the tiny rectifier diodes and the health of the switch though...
I’ve actually got this irrational urge to hot glue the original components to the steel chassis, after replacement, for future owners to marvel at...
The main PS filter cap is on the left, 1500uF at 63v and the two output caps on the right.
I could get a 10,000uF 100v PS filter cap in there at the same diameter and to the height of the case and who knows what I’m the output cap section.
I’ve limited myself to 4700uF for the PS in deference to the tiny rectifier diodes and the health of the switch though...
Attachments
Only if the speaker’s impedance is pure capacitive, which it is not. It is only capacitive in some portion of the band (below fundamental resonance, and sometimes just below the crossover frequency), and there is always some real part which will limit the VI product on the transistor. The turn on transient won’t be any harder on the transistors than driving the load normally would be.
It’s only a 52 volt single power supply. +/-26 in today’s world. With 2SD180’s in one channel, and 1972-vintage 2N3055’s in the other, according to the pic. Both are going to be the old bulletproof hometaxial (single diffused) types, limited by VCEO not SOA. They’d use TIP41/42 today.
The decision as to what value output AND power supply cap to use comes down to how much low bass you can get out of the speaker. It’s a 20 Hz 3dB point, 40 Hz 1dB point, and 100 Hz 0.1 dB point. The power supply cap value is tied to the speaker coupling cap. It needs to be at least the same as the output cap value (double or more if using split supplies, because those two caps are in series). If you up-size the output cap, the power supply cap should follow in proportion. Bass will not clip cleanly if you don’t. If you use 4700 in the supply, limit to 4700 for the output cap. Preferably one size smaller - say 3300. That’s what they did in the original design, 1500 vs. 1000.
It’s only a 52 volt single power supply. +/-26 in today’s world. With 2SD180’s in one channel, and 1972-vintage 2N3055’s in the other, according to the pic. Both are going to be the old bulletproof hometaxial (single diffused) types, limited by VCEO not SOA. They’d use TIP41/42 today.
The decision as to what value output AND power supply cap to use comes down to how much low bass you can get out of the speaker. It’s a 20 Hz 3dB point, 40 Hz 1dB point, and 100 Hz 0.1 dB point. The power supply cap value is tied to the speaker coupling cap. It needs to be at least the same as the output cap value (double or more if using split supplies, because those two caps are in series). If you up-size the output cap, the power supply cap should follow in proportion. Bass will not clip cleanly if you don’t. If you use 4700 in the supply, limit to 4700 for the output cap. Preferably one size smaller - say 3300. That’s what they did in the original design, 1500 vs. 1000.
No fear, all the space in the world.
I’ve actually got this irrational urge to hot glue the original components to the steel chassis, after replacement, for future owners to marvel at...
The main PS filter cap is on the left, 1500uF at 63v and the two output caps on the right.
I could get a 10,000uF 100v PS filter cap in there at the same diameter and to the height of the case and who knows what I’m the output cap section.
I’ve limited myself to 4700uF for the PS in deference to the tiny rectifier diodes and the health of the switch though...
What's with the butchering going on there?
I mean that PC fan.
That chassis doesn't need that dumb thing in there!
Take it out and let the thing work the way it should.
If needed, replace the output coupling caps with 1000/50v, and the PS cap with 4700U 75v
Adjust center V if possible to get 25.5v on the hot side of the output caps and call it a day.
Last edited:
I have worried about the output transistors, to be honest. The 2SD180s are listed as original in one of the schematics available from Hifiengine (It's a Sansui 350) but my chances of finding a pair of them new are very low. They do appear to be faster (Ft ~ 10Mhz) and have a lower Cob (150pf) compared to the vintage 2N3055s (probably around 400Khz and 1000pf) and I have wondered whether they are making one channel sound slightly "fuzzy" in the treble but have also considered that it might be all in my imagination...
I can't seem to find a readily available substitute, any recommendations would be gratefully accepted.
I ended up using a 4700uF Nichicon 105C job for the main filter cap, it handily fit in the holes drilled in the PCB for the original 4 pin 1500uF cap and despite the pin spacing being half that of the original, still cleared other componentry.
For the output I used low ESR 2200uF 50V jobs and recapped the rest of the tone and power boards with ordinary 105C Chinese caps from a ebay kit.
The idea is to listen thoroughly and then replace with Elna Silmics, Panasonic FRs, films for small values, etc to see if there's a difference. I was thinking of replacing the carbon resistors in the signal path and feedback loop with metal film ones at some point, in the same vein.
The amplifier is so easy to work on, compared to most and especially alter monster receivers, that it lends itself to this type of experimentation.
It's not double blind, A/B testing but fun and something to look forward too.
The amplifier always had surprising bass for such a dinky thing, if not pushed too hard.
The bass response from the phono input seems much fuller and warmer than the aux in from my DAC, perhaps the RIAA curve has drifted and needs attention. The Phono section in general seems to need some help, it's noisy and there is some definite hum. I didn't recap the power supply for the preamp ( a capacitor multiplier), I should probably start there and then look at replacing transistors to find the source of the low level hiss.
The fan was just to dry the PCB after cleaning, the use some sort of 'orrible conformal coating that smells like cat **** when it gets warm.
The caps have been replaced and the amp will need to be listened to for a while to see if any improvements have actually been made as a result of the whole process.
Initial listening impressions are good but I'm researching exactly that bias setting procedure as a sine wave clips at the bottom first and the service manual for a Sansui 350 from HifiEngine doesn't cover the adjustment procedure. I found some information to the effect of setting the DC bias to 17mA +-2mA, Presumably by measurring across the 0.5 Ohm output resistors but am a little hay on how to go about it and adjusting the "centre V", as you put it.
Thinking about it again, of course, it might actually be cat ****...
Yes, for some large value of "way too big".
Actually, with modern cap prices, you could be in trouble just by "matching physical size" of a 50 year old cap.
The speaker impedance hardly matters. In most cases it is always higher than DCR. It does go capacitive for an octave above resonance, but typically hundreds of uFd, which just makes the output cap look smaller. (1000uFd+300uFd= 231uFd) Going cappy on the high side of resonance may be an efficient "thump".
The cap has a peak current of V(peak)/R, as expected. Just under 2A (20V/10Ω). For just-big-enuff 1,000uFd this lasts for about 10mS. This is way up the SOAR graph for any large part. For gross-oversize 10,000uFd the surge is OTOO 100mS, which for most of our parts is in the "maybe" zone.
Since 2Hz response on a 20 Watt amp is pointless, it seems wise to stick to say 1,000-3,300uFd, not ten thousand.
In any real AC-coupled amplifier the center-bias filtering will massively slow-down this charging spike. Not to zero but perhaps little danger of crossing SOAR.
In this simulation I used the "Skip initial..." option so it starts cold, not pre-charged like a normal simulation. (I tried a thing with a time-switch but the sim barfed it back.)
Attachments
Last edited:
This is worrying, no way the amplifier would have mixed these two in production. The 2SD180 will be a lot more fragile than an old 2N3055, but give far better cross over distortion. The question is which part was the amplifier compensation designed for?
The BD180s, certainly, they were the originals.
I have a HP 333A that I am hoping to be able measure the distortion (and look at the distortion products) of the two channels to see if there is much difference between the channels , but was waiting until I had set the centre point and bias correctly.
The thing is noisy, now that I have got it into the main system and listened to it properly. A hiss and occasional sounds like distant popcorn...
In terms of transistors if I have on hand from a helpful old boy with a stock of NOS are BDN21B and 2N5873.
They’re not quite up to the BD180 but should be faster than the 2n3055 in there now, what do we think?
I have a HP 333A that I am hoping to be able measure the distortion (and look at the distortion products) of the two channels to see if there is much difference between the channels , but was waiting until I had set the centre point and bias correctly.
The thing is noisy, now that I have got it into the main system and listened to it properly. A hiss and occasional sounds like distant popcorn...
In terms of transistors if I have on hand from a helpful old boy with a stock of NOS are BDN21B and 2N5873.
They’re not quite up to the BD180 but should be faster than the 2n3055 in there now, what do we think?
Looks like an ancient metal can (TO-1?) transistor where I suppose an input stage is. Could even be Ge - that case was used on types like the 2SB56. I wouldn’t put it past the small signal transistors of the era to sound like popcorn. Like the old infamous 2SA720 and 2SC458 in that horrible air-cavity TO-92. Every single one of those I’ve seen goes bad and sounds like popcorn or distant thunder. Old resistors can get noisy too, so don’t discount those either.
There is probably nothing *wrong* with those old 3055’s, but a ruggedized device isn’t really needed here unless you use that for short circuit protection. I might pull them just to use them where they would be uniquely suitable, but if it’s working you could just leave it alone. Those old 2N587x types are more “audio” suitable - especially on a 52 volt supply. I have seen gain values all over the map on those - if you have a bunch to choose from you may want to match up a pair, and select for a gain of say 70. The PNPs will get into the 200 range at low current. Those old D180’s aren’t exactly what I’d call “fast” either. Nothing like today’s sustained-beta flatpacks. Again, nothing wrong with them in the application. Today’s amps would be using TO-220’s and really have a problem with burnout.
There is probably nothing *wrong* with those old 3055’s, but a ruggedized device isn’t really needed here unless you use that for short circuit protection. I might pull them just to use them where they would be uniquely suitable, but if it’s working you could just leave it alone. Those old 2N587x types are more “audio” suitable - especially on a 52 volt supply. I have seen gain values all over the map on those - if you have a bunch to choose from you may want to match up a pair, and select for a gain of say 70. The PNPs will get into the 200 range at low current. Those old D180’s aren’t exactly what I’d call “fast” either. Nothing like today’s sustained-beta flatpacks. Again, nothing wrong with them in the application. Today’s amps would be using TO-220’s and really have a problem with burnout.